plan-before-code
/plan-before-code
Decide the shape and the size of a change before writing it. Use for any non-trivial edit, a new mechanism, a refactor, or work whose scope is not obvious from the request.
- Author
- alexboo86 ↗
- Repository
- booharness ↗
- Licence
- MIT
- Last repository change
- 02.11.2026
Assessment and security
- Files in this package
- 1
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 02.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
What it does and when to use it
Decide the shape and the size of a change before writing it. Use for any non-trivial edit, a new mechanism, a refactor, or work whose scope is not obvious from the request.
Original author description
Decide the shape and the size of a change before writing it. Use for any non-trivial edit, a new mechanism, a refactor, or work whose scope is not obvious from the request.
How it works
SKILL.md has no level-two headings · 20 lines, 276 words
SKILL.md preview
--- name: plan-before-code description: Decide the shape and the size of a change before writing it. Use for any non-trivial edit, a new mechanism, a refactor, or work whose scope is not obvious from the request. --- # Plan before code Name the cause and the smallest change that removes it, before the first line. - State what breaks today and what is true once it is fixed. - Walk the ladder: needed at all, already here, standard library, platform, an installed dependency, one line, then the minimum that works. - Name what the change touches and what must keep holding afterwards. - Name the touched module ids with `node scripts/module-registry.mjs --out <file>` or `docs/contracts/module-index.md` (not in this kit). Read each contract's Public API and Invariants. State which invariant the change keeps or changes. A Public API change updates its contract in the same commit. - Name the assumption whose falsification would change the plan. - Ask the plan its blind spots: who may do this and what if they may not; empty, one and maximum; a dependency down; two actors at once; related data left consistent; invalid or duplicate input; whether it can be undone; the volume it must hold. - Decide the size BEFORE writing: a larger design is the operator's money, so say its price out loud. - Stop planning when the next step is obvious and its proof is named. A lead's plan goes to a [plan critic](../plan-critic/SKILL.md) before the first coder only when a wrong plan costs more than one correction turn; the five cases are in [orch](../orch/SKILL.md), "The loop" (operator #15067). Every other plan has the reviewer of the diff as its check. Report the plan and the check that decides it. Write no code the plan does not require. Adapted in part from addyosmani/agent-skills (MIT), read 2026-09-23; the blind-spot list from Habr 1085648 (Mindbox), read 2026-10-01.
Versions
| Repository | Score | Licence |
|---|---|---|
| This page · alexboo86/booharness | 86 | MIT |
| ThalesGroup/agilab | 100 | BSD-3-Clause |
Database snapshot: 13.11.2026 · package version 9b6d4a067438 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 96pr
The shape of a pull request body. Use when writing or editing a PR description. - 90plan-feature
The Explore to Plan to Code to Commit workflow for any non-trivial change. Forces understanding and an explicit plan before edits, with a one-sentence skip heuristic for trivial work. Use when starting a feature, a refactor, or any change whose steps are not obvious. Pairs with plan mode. - 100plan-feature
Use when the user says "plan a feature", "design this", "how should we build", or before writing code for any non-trivial feature. Produces an approved task breakdown before implementation begins — plan first, build second. - 100modularity
This skill should be used when designing, planning, implementing, or reviewing any non-trivial change, or when the user asks to "split this file", "decouple X from Y", "refactor into modules" — enforces small, focused modules so any change fits in context without compaction - 90architect
Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in. Use for /architect, 'architect this', 'design this', or non-trivial work where jumping to code would lock in the wrong shape. - 90architect
Sketch types, signatures, and module structure before code, then stay in the loop while implementation fills in. Use for /architect, 'architect this', 'design this', or non-trivial work where jumping to code would lock in the wrong shape.